An agentic AI deployment will cause a publicly disclosed data breach next year, leading to employee dismissals, Forrester has predicted.
Senior analyst Paddy Harrington noted that generative AI (GenAI) has already been responsible for several breaches since it burst onto the scene three years ago.
“As companies begin building agentic AI workflows, these issues will only become more prevalent,” he added in a blog post yesterday.
“Without the right guardrails, systems of autonomous AI agents may sacrifice accuracy for speed of delivery, especially when interacting directly with customers.”
Although staff may lose their jobs as a result of an agentic AI-related breach, this would be unfair given that such events are due to “a cascade of failures,” rather than the fault of individuals, Harrington continued.
“To prevent these failures, and scapegoating, security organizations must enable the business to develop agentic applications with minimum viable security,” he added.
“Follow the AEGIS framework, securing intent, ensuring appropriate identity and access management controls to track agent activity, and implementing data security controls to track data provenance.”
Forrester’s Agentic AI Enterprise Guardrails for Information Security (AEGIS) framework focuses on six core elements
- Governance, risk, and compliance (GRC)
- Identity and access management (IAM)
- Data security and privacy
- Application security
- Threat management
- Zero Trust architecture
Read more on agentic AI security threats: OWASP Launches Agentic AI Security Guidance
Forrester isn’t the only analyst house to raise the alarm over possible agentic AI security risks. Back in March, Gartner predicted that within two years, AI agents will accelerate the time it takes threat actors to hijack exposed accounts by 50%.
Agentic AI Already Helping Threat Actors
Although Forrester’s prediction is based on a scenario where enterprise AI agents accidentally or deliberately breach data, the technology is already being used maliciously, according to Check Point.
The security vendor claimed last month that the agentic AI-powered Hexstrike-AI red team tool is being abused by threat actors to accelerate reconnaissance, exploit crafting and payload delivery.
It warned that a task which could have taken days or weeks can now be accomplished in under 10 minutes.
Among Forrester’s other cybersecurity predictions for 2026 are:
- Quantum security spending will exceed 5% of overall cybersecurity budgets, as enterprises seek to plan their post-quantum migration, replace outdated cryptographic libraries and components, and invest in discovery and inventory tools
- At least five governments will nationalize or place restrictions on critical telecoms infrastructure, following the Salt Typhoon attacks
- The EU will launch its own known exploited vulnerability (KEV) database, as CISA funding wanes, based on the EU Vulnerability Database (EUVD). This will likely require CISOs to adjust their vulnerability management strategy